Real Time Face Tracking and Recognition Using Efficient Face Descriptor and Features Extraction Algorithms

Abstract

The real time face recognition system used to capture the real time face of end user, detect the face, and perform the recognition by preexisting features of the corresponding user. The real time face tracking and recognition is a challenging research problem since last two decades due to the various presence of various illumination conditions, low-resolutions face images, different facial expressions, etc. in capture face images. In this research work, we proposed a novel framework for the real time face tracking and recognition regardless of input facial research dataset. The proposed method is composed of four main steps such as pre-processing, face descriptor, features extraction and selection, and classification. In the first phase, we first perform the task of face tracking and cropping the detected face using the Viola–Jones (VJ) method, we further applied the Gaussian filtering to smooth the cropped face image. In the second phase, the novel face descriptor method called Tracked Directional Ternary Pattern (TDTP) was proposed to address the real time face recognition challenges. In features extraction method, we designed a novel modified DCT (MDCT) based feature extraction method to address the challenge of low-resolution face images.